Browse All Artificial Intelligence Content (631)

Heidi Hämäläinen explains why Microsoft Purview Data Governance can feel heavy at first, and why governed metadata (glossary, catalog, data products, and security foundations) matters for scalable analytics, ML, and GenAI work—especially when you need discoverability, compliance, and trust in production.

OsvaldoDaibert explains a common IBM Power → Azure x86 migration failure mode in C++: silent integer corruption caused by Big-Endian vs Little-Endian byte order, and shows a practical refactoring workflow—accelerated with GitHub Copilot—to add portable byte-swapping, guard against struct padding, and deploy to Azure.

aycabas announces the IQ Series: Foundry IQ, a set of developer-focused episodes (with videos, Jupyter notebooks, and a GitHub repo) that walks through building knowledge-centric AI systems where agents query structured knowledge bases across sources like Azure AI Search, Blob Storage, Fabric/OneLake, SharePoint, and the web.

In this NVIDIA GTC update, stclarke outlines Microsoft’s announcements across Microsoft Foundry and Azure: Foundry Agent Service GA with control-plane observability, voice agent preview capabilities, expanded model access (including NVIDIA Nemotron), and new Azure AI infrastructure plus “Physical AI” tooling that connects simulation, data, and real-world operations.

Bala Venkataraman, jeffhollan, and Nick Brady announce the GA release of Foundry Agent Service, highlighting enterprise features like private networking, expanded MCP authentication options, Voice Live speech-to-speech integration, and built-in evaluations with continuous monitoring via Azure Monitor.

meenagowdar explains how Azure Local can support “sovereign AI” by running advanced model workloads inside customer-controlled environments, from current NVIDIA RTX PRO 6000 Blackwell support (Azure Local 2603) to planned NVIDIA Rubin support, with Foundry Local services, AKS on Azure Local, and Azure Arc for management and governance.

Microsoft Developer demonstrates an end-to-end workflow for building an agentic AI solution with the VS Code AI Toolkit and Microsoft (Azure AI) Foundry—from model selection and prototyping to hosting, deployment, evaluation, red teaming, and monitoring.

Darren Portillo outlines Microsoft Purview updates for Microsoft Fabric focused on preventing data oversharing and improving governance and data quality, with new DLP, Insider Risk Management, DSPM, and Unified Catalog capabilities aimed at supporting safer AI adoption.

GitHub explains how the .github folder can standardize repository workflows—covering CI/CD automation, issue templates, and even rules that influence how GitHub Copilot behaves across projects in an organization.

Matt Soucoup announces a major update to the Awesome GitHub Copilot Customizations project: a new website, a Learning Hub, and a plugin system to make Copilot agents, skills, instructions, and automation easier to find, install, and contribute to.

Welcome to this week’s roundup. The common thread is agents moving beyond “helpful chat” into real execution across IDEs, terminals, CI, and cloud operations. Copilot’s latest changes focus on autonomy and repeatable behavior through repo-visible instruction files, lifecycle hooks, clearer model routing, and faster PR review workflows, while modernization tooling ties assessments and plans directly to issues and pull requests. In parallel, the rest of the stack is catching up to the day-to-day requirements of running agents like software: traces and debugging loops, structured outputs and schema enforcement, and clearer guardrails around approvals, secrets, and identity-based access.

Visual Studio Code shares a walkthrough with Justin and James on VS Code’s evolving chat UX and Autopilot (Insiders preview), covering how agent-style workflows can auto-approve tools, iterate toward a task_complete signal, and how the new permissions picker and input bar changes affect safe, hands-off usage.

carlottacaste walks through an end-to-end workflow for taking an agent from prototype to production using the AI Toolkit in VS Code and Microsoft Foundry, covering model selection, agent setup, migration to hosted code, deployment, and ongoing evaluation/monitoring.

GitHub shows how to use GitHub Copilot to help write code for a robot dog project, generating a custom servo “greeting” sequence that triggers when a Raspberry Pi 5 camera detects a human face.

Satya Nadella shares a milestone for Microsoft, leading the cloud market by validating the NVIDIA Vera Rubin NVL72 system—a foundational building block for future AI infrastructure.

Diego Casati and Ray Kao explore the challenges of scaling platform engineering processes, focusing on specialization, tool complexity, and the growing role of AI-powered collaboration in the Azure ecosystem.

PuiChee (PC) Chan explains how to use new Azure Developer CLI (azd) commands to diagnose and debug hosted AI agent failures, providing developers with direct access to container status and live logs from the terminal.

Allison announces a new option in GitHub that lets repository administrators skip manual approval for GitHub Actions workflows triggered by Copilot coding agent, balancing workflow speed and security.

stclarke outlines Microsoft's new systematic debugging approach for AI agents, introducing the AgentRx framework to improve AI agent failure analysis and reliability.

NagaSurendran details practical strategies for organizations migrating from Heroku, focusing on how Azure and its integrated tools—including GitHub Copilot and Microsoft Foundry—enable modern, secure, and intelligent cloud-native applications.